Success and Failure Suppressing Reflexive Behavior

Clayton E. Curtis and Mark D’Esposito Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 Abstract & The dynamic interplay between reflexive and controlled pre- before voluntary antisaccades determinants of behavior is one of the most general organizing (saccades away from a target) compared with reflexive principles of brain function. A powerful analogue of this prosaccades (saccades to a target). Moreover, this preparatory interplay is seen in the antisaccade task, which pits reflexive activity was critically associated with reflex suppression; it and willed saccadic mechanisms against one another. Event- predicted whether the reflex was later successfully inhibited in related functional magnetic resonance imaging of the human the trial. These data illustrate a mechanism for top-down control brain showed greater prestimulus preparatory activity in the over reflexive behavior. &

INTRODUCTION successful suppression of saccades during antisaccade We often perform behaviors reflexively in response to trials (Everling, Dorris, & Munoz, 1998; Everling, Dorris, our environment. For instance, the sudden appearance Klein, & Munoz, 1999). It is thought that top-down of a visual stimulus captures our attention automati- controlling projections from cortical regions to the SC cally (Yantis & Jonides, 1984) causing our gaze to can suppress reflexive saccade generation. Neurophysio- reflexively shift to the location of the stimulus. How- logical studies of monkeys performing antisaccades ever, our gaze is not solely visually guided and in identify several cortical regions as likely sources of top- general we are capable of using internal motives to down oculomotor control. Neurons in the lateral intra- bias behavior against strong externally triggered parietal (LIP) region are thought to mainly convey task reflexes (Miller & Cohen, 2001). Willful control pro- relevant visual information, but to a much lesser extent cesses can override a reflex if a cancellation signal can saccade metrics, to the saccadic system through its be issued in time (Hanes, Patterson, & Schall, 1998). projection to the SC (Gottlieb & Goldberg, 1999). The source and nature of these control processes are Importantly, two premotor regions of the frontal cor- key to understanding behavior but remain largely tex—the supplementary eye fields (SEF) (Schlag & unknown.The antisaccade task (Hallett, 1978) is an Schlag-Rey, 1987), which reside within the rostralmost experimental analogue of the dynamic interplay extent of the supplementary motor area (SMA), and the between reflexive and controlled determinants of (FEF) (Paus, 1996; Bruce, Goldberg, behavior; it requires willful inhibition of a powerful Bushnell, & Stanton, 1985), which reside in the dorsal drive to reflexively saccade to an abrupt visual stimulus. precentral sulcus near its junction with the superior The sudden appearance of the stimulus, which height- frontal sulcus—show single-unit activity predictive of ens its saliency (Yantis & Jonides, 1984), leads to the successful suppression of unwanted saccades that pre- prepotency of the reflexive saccade. Conflict between cedes the appearance of the saccade stimulus (Schlag- reflexive and controlled saccadic processes arises from Rey, Amador, Sanchez, & Schlag, 1997). the need to use the location of stimulus to guide the These frontal premotor regions, medial and lateral, direction of the saccade, but to do so without looking have been implicated in general antisaccade perfor- at it. mance in humans as well (Kimmig et al., 2001; Connolly, The superior colliculus (SC) in the midbrain primarily Goodale, Desouza, Menon, & Vilis, 2000; Sweeney et al., mediates reflexive orienting of gaze to an abruptly 1996; O’Driscoll et al., 1995). The medial wall in area 6 presented visual stimulus (Dorris, Pare, & Munoz, is the substrate for two motor regions whose separation 1997; Schiller, Sandell, & Maunsell, 1987). In the mon- is supported by anatomical connectivity, cytoarchitec- key, prestimulus activity (i.e., activity before the onset of tonic, and functional differences (Picard & Strick, 2001; the peripheral saccade stimulus) in the SC determines Rizzolatti & Luppino, 2001) and whose division is roughly marked by the vertical plane of the anterior commissure (VAC). The SMA proper, including the SEF, University of California lies caudal to the VAC, while pre-SMA sits rostral to the

D 2003 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 15:3, pp. 409–418

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 Figure 1. Saccade stimuli and eye-position acquisition. (a) Schematic depiction of a trial. Each trial began with a fixation dot that briefly changed colors instructing the participant to make a prosaccade (green) or antisaccade (yellow). The instruction was followed by a 6-sec delay and a 200-msec gap, after which the peripheral saccade stimulus appeared in Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 one of eight radial positions (68 from center). (b) Examples of eight visually guided prosac- cades made to targets recorded in the scanner. (c) Examples of four reflexive saccade errors made during antisaccade trials. Each shade and line thickness represents a different trial. Notice the prototypical pattern of a quick reflexive saccade made towards the target followed by a corrective antisaccade away from the target. Participants made reflexive errors on 13.1 ± 6.9% of antisaccade trials.

VAC. Importantly, investigations have observed greater Sweeney et al., 1996; O’Driscoll et al., 1995) subtracted activity in the SMA proper during the execution of activity summed over long blocks of antisaccades, relatively simple motor acts while greater activity has whether correct or not, from activity over blocks of been found in the pre-SMA during motor responses prosaccades, underutilizing the valuable temporal infor- guided by more demanding sensory–motor transforma- mation available in fMRI. We used, instead, an event- tions (Rushworth, Hadland, Paus, & Sipila, 2002; Gros- related design to capitalize on the temporal resolution bras et al., 2001; Kurata, Tsuji, Naraki, Seino, & Abe, of MR signals, allowing us to examine activity specifi- 2000; Deiber, Honda, Ibanez, Sadato, & Hallett, 1999; cally from the trial period before the onset of the Sakai et al., 1999; Kawashima et al., 1998; Petit, Court- peripheral stimulus (Figure 1a). Critically, activity in ney, Ungerleider, & Haxby, 1998; Hikosaka et al., 1996). this interval reflects preparatory processes that are Oculomotor control was studied by Merriam et al. separated in time and are thus uncontaminated by (2001), who found greater SEF activity during blocks saccade production processes. In addition, we recorded of endogenously cued (i.e., with the central presenta- eye position in the scanner during functional imaging. tion of the words LEFT or RIGHT ) saccades and greater This allowed us to later sort and compare the fMRI pre-SMA during endogenously cued antisaccades. These data by performance (Figure 1b,c). Together, these data suggest that greater oculomotor control is associ- methodological advances finally permit testing of two ated with increased utilization of the pre-SMA. However, predictions that arise from the proposed brain–behavior other recent human functional magnetic resonance relationship: (a) Prestimulus activity is greater prior to imaging (fMRI) studies of antisaccade performance have antisaccades than prosaccades, reflecting the need for not supported this functional separation of pre-SMA and greater control; (b) This early prestimulus difference is SEF (Kimmig et al., 2001; Connolly et al., 2000). indeed critical to the success or failure of saccade Using event-related fMRI of antisaccade performance suppression later in the trial. By requiring both greater at high field (4 T), we test whether the pattern of brain activity when inhibitory control is needed and disinhibi- activity in the human pre-SMA, SEF, FEF, and intra- tion or failed suppression when activity is less, these two parietal sulcus (IPS) relates to the heightened need for hypotheses form a strict test of the proposed brain– behavioral control over reflexive orienting of gaze. The behavior relationship. In this report, we focus on the previous neuroimaging studies to date (Kimmig et al., SEF, pre-SMA, FEF, and IPS (Gottlieb & Goldberg, 1999) 2001; Merriam et al., 2001; Connolly et al., 2000; as the potential sources of top-down control over

410 Journal of Cognitive Neuroscience Volume 15, Number 3

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 reflexive saccades. We presented these data in abstract and prosaccade trials and the difference between anti- form previously (Curtis & D’Esposito, 2001). saccade trials where the initial reflexive prosaccade was successfully and unsuccessfully inhibited. RESULTS Instructional Cue Period All regions of interest (SEF, pre-SMA, FEF, and IPS) showed significant Trial Type (prosaccade, antisaccade Follow-up t tests failed to find any significant differences correct, and antisaccade error) Â Trial Period (instruc- in any region between any trial types during the instruc- tional cue, delay, response) interactions (all ps < .05). tional cue period. Thus, although each region was Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 Thus, differences between conditions at each trial peri- responsive to the instructional cue (relative to the od are presented below. Again, the crucial test for each intertrial interval), there were no significant differences region-of-interest is the difference between antisaccade among the trial types.

Figure 2. Group activations along the medial wall of the frontal cortex in the pre-SMA and the SEF. Preparatory delay period activations are depicted in (a) and (c), while stimulus-response period activations are represented in (b) and (d). Greater correct antisaccade compared with prosaccade activations are depicted in (a) and (b), while greater activations during correct antisaccades compared with reflexive prosaccade errors on antisaccade trials are depicted in (c) and (d). The pre-SMA shows greater preparatory delay period activity when subjects are anticipating the need to inhibit the reflexive prosaccade and make an antisaccade (a) and when subjects fail to suppress the reflex, activity is diminished (c). The SEF does not show greater antisaccade than prosaccade activity during the preparatory delay (a), however, it does show greater delay period activity when comparing correct antisaccades to reflexive prosaccade errors during the delay (c) and the SEF shows strong differential effects at the response period of the trial (b and d). Average time series data from the (e) pre-SMA and (f ) SEF regions showing activity during prosaccade trials and antisaccade trials when the reflexive saccade was success- fully and unsuccessfully suppressed. The gradated color bar at the bottom approxi- mately references from which epoch of the trial the signal is emanating given the 4- to 6-sec lag in the hemodynamic response.

Curtis and D’Esposito 411

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 Preparatory Delay Period showed significantly greater activation during the pre- paratory delay interval for both comparisons: correct Beginning shortly after participants were given the anti- antisaccade versus prosaccade and correct antisaccades saccade instruction, activity in the pre-SMA began to versus antisaccade errors. increase and significantly diverge from the activity seen To confirm the spatial representativeness of the group- during prosaccade trials, t(11) = 6.01, p < .00007. Note level activation in the pre-SMA shown in Figure 2a, that this difference in the pre-SMA, when subjects the location of each of the 12 subject’s delay period anticipated an increased need for oculomotor control, activity is plotted separately in Figure 3. As can be seen, occurred during the preparatory delay interval (i.e., all 12 subjects showed robust delay period activity along 2–4 sec prior to the presentation of the peripheral Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 the medial frontal wall that was greater before antisac- saccade stimulus that cued where they were to shift their cades compared with prosaccades. Importantly, 9 of the gaze) (Figure 2a). Moreover, on trials when participants 12 subjects showed activations that peaked rostral to the failed to successfully suppress the reflexive saccade, VAC plane in the pre-SMA. Only two subjects had activity in the pre-SMA was significantly lower during activations that were limited to an area that is consistent the preparatory delay interval compared with trials when with the SEF, caudal to the VAC in or near the para- they were successful, t(11) = 6.31, p < .00006 (Figure 2c). central sulcus. Inspection of the averaged time-courses from the 12 participants highlights the important role that the pre- SMA plays in oculomotor control (Figure 2e). Stimulus-Response Period Neither the SEF, t(11) = 2.86, p > .01, right FEF, t(11) = 2.76, p > .01, left FEF, t(11) = 2.69, p > .01, All regions of interest showed significantly greater acti- right IPS, t(11) = 2.05, p > .06, nor the left IPS, t(11) = vation at the response period when the execution of 1.92, p > .08, showed significantly greater activity dur- correct antisaccades was compared with the execution ing the preparatory delay period for correct antisaccade of prosaccades and compared with instances when trials compared with prosaccade trials. However, all of reflexive errors were made on antisaccade trials (all these regions, with the exception of the left IPS, t(11) = ts > 4.11 and all ps < .002). Particularly strong signal 2.83, p > .01, showed greater activation in the delay differences were found in the FEF [antisaccade > pro- interval prior to correct antisaccades compared with saccade, t(11) = 6.44, p > .00005; antisaccade correct > reflexive errors made on antisaccade trials (all ps< antisaccade error, t(11) = 6.35, p > .00006] and the IPS .008). Thus, the pre-SMA was the only region that [antisaccade > prosaccade, t(11) = 10.79, p > .0000004;

Figure 3. Individual subject activations along the medial wall of the frontal cortex illustrating greater preparatory delay interval for antisaccade compared with prosaccade trials. The vertical red lines indicate the vertical anterior commissure (VAC) line. Note that in 9/12 subjects (S1–S9), there is greater preparatory delay activity just rostral to the VAC. Numbers in the upper right-hand corner indicate the y-coordinate of the peak activation in millimeters relative to the VAC. In two subjects (S10 and S11), the activity is limited to a region just behind the VAC about 1–2 mm. In S12, the only activity seen is in the cingulate gyrus, not the SEF or pre-SMA.

412 Journal of Cognitive Neuroscience Volume 15, Number 3

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 Figure 4. The FEF and IPS, bilaterally, demonstrate increased activation that is limited to the stimulus- response period of the trial during the execution of correct antisaccades compared with (a) the execution of prosaccades and (b) when subjects failed to suppress the reflexive prosaccade. Group statistical parametric maps are Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 overlaid on a single subject’s anatomy. Average time series data from the (c) FEF and (d) IPS showing activity during prosaccade trials and antisac- cade trials when the reflexive saccade was successfully and unsuccessfully suppressed. Time series are aligned to the onset of the peripheral stimulus prompting an antisaccade or prosaccade.

antisaccade correct > antisaccade error, t(11) = 8.91, presentation of the stimulus that cues the direction of p > .000002] regions for each contrast (Figures 2 and 4). the saccade likely reflects processes relating to super- visory control that facilitate appropriate motor behavior when needed, in this case when participants were Response Profile Analysis anticipating the need to inhibit the visual grasp reflex. In order to compare the differential activations of On trials when participants subsequently failed to suc- regions across the three periods, effect sizes (Cohen’s d ) cessfully inhibit the reflexive saccade, and presumably for the main contrasts were computed. As can be seen in failed to sufficiently activate at least one node of the Figure 5, the response profiles of the pre-SMA and SEF circuit responsible for inhibitory saccadic control, the look qualitatively different than that of the FEF and IPS. activity in the pre-SMA was reduced. Thus, pre-SMA The pre-SMA and SEF show the largest effect during the activity meets both of the requirements we listed above delay period when subjects successfully prepare to linking its activity to behavior. Namely, activity during inhibit a saccade, while the FEF and IPS show the largest the preparatory delay interval increases when inhibitory effect at the response period. These different response control is anticipated, and the level of this activity is profiles suggest different roles in oculomotor control. critically associated with the successful withholding of unwanted saccades. No other region met these two strict requirements. DISCUSSION Our findings support the view that the pre-SMA is These human data show increased prestimulus, prepar- functionally distinct from other frontal premotor atory delay period activity in pre-SMA before antisac- regions, including the SMA proper (Picard & Strick, cades compared with prosaccades. Beginning shortly 2001; Rizzolatti & Luppino, 2001). Among the premotor after participants were given the antisaccade instruc- regions, the pre-SMA is situated to play a higher level tion, activity in the pre-SMA began to increase and role in the perception–action hierarchy and may oper- diverge from the activity seen during prosaccade trials. ate more like a heteromodal cortical association area This increase in pre-SMA activity 2–4 sec prior to the (i.e., like prefrontal cortex) than a unimodal cortical

Curtis and D’Esposito 413

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 Hopfinger, 2000; Garavan, Ross, & Stein, 1999; Humber- stone et al., 1997) and the updating or switching of essential visual–motor associations (Rushworth et al., 2002; Grosbras et al., 2001; Heide et al., 2001; Sakai et al., 1999; Kawashima et al., 1998; Shima, Mushiake, Saito, & Tanji, 1996). Consistent with these data, the observed pre-SMA activity during the preparatory delay interval in the current study may reflect the anticipated need for inhibition of the eminent programming of a Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 reflexive, visually guided saccade. However, the pre-SMA may not necessarily be responsible for actually inhibiting the reflexive response. Instead, we propose that the pre- SMA readies or prepares other oculomotor regions, like the SEF, in some fashion such that reflexive responding is less likely. Of course, temporary maintenance of the instructional cue was required for successful saccade performance; the instruction appeared only briefly at the beginning of the trial, 6 sec before the correct saccadic response could be selected. Thus, it could be argued that the greater activity in the pre-SMA was due to the main- tenance of the instructional cue across the delay period (i.e., retrospective memory code) (Petit et al., 1998). However, it does not seem likely that remembering an instruction to later perform a prosaccade (‘‘move eyes towards the stimulus’’) versus remembering an instruc- tion to later perform an antisaccade (‘‘move eyes away from the stimulus’’) should place a different load upon working memory systems. If this were the case, then one might predict a greater number of responses consistent with forgetting on antisaccade compared with prosac- cade trials. In other words, one might expect more prosaccades not followed by corrective antisaccades (i.e., not a reflexive error) during instructed antisaccade trials than antisaccade errors made during instructed prosaccade trials. This was not the case. Rarely (i.e., less Figure 5. Profile plots summarizing the trial period by region effects. than 0.6% of all trials for all subjects) did subjects make Each element represents the effect size for the region-of-interest the wrong and uncorrected eye movement that would during a given trial period, where the effect size (d ) is equal to the be expected if they had forgotten the instructional cue; mean parameter estimate for a given contrast divided by the group these rare instances did not occur disproportionately on standard deviation. Effect sizes for (a) antisaccades greater than prosaccades and (b) correct antisaccades greater than errors on antisaccade trials. Although this observation does not antisaccade trials. See text for details. exclude the possibility, it is inconsistent with the inter- pretation that the difference in delay period activity represents a retrospective memory code. area (i.e., like premotor cortex). Anatomical and func- More likely, the differential processing demands dur- tional evidence in human and nonhuman primates ing the delay interval on different trial types represent a supports this claim. First, the monkey pre-SMA is much prospective memory code or a preparatory motor set more interconnected with the prefrontal cortex than that would be greater for an upcoming antisaccade the SMA proper (Lu, Preston, & Strick, 1994; Bates & compared with a prosaccade given the additional oper- Goldman-Rakic, 1993; Luppino, Matelli, Camarda, & ations required to generate an antisaccade. These addi- Rizzolatti, 1993). Second, pre-SMA activity has been tional operations include the inhibition of the visually implicated in higher level cognitive control rather than evoked saccade, the 1808 transformation of the saccadic motor behavior. The pre-SMA has shown consistent vector, and finally the execution of the antisaccade itself. sensitivity to response inhibition during go/no-go and In this light, the preparatory activity in the pre-SMA flanker tasks (Menon, Adleman, White, Glover, & Reiss, might be the source of the Bereitschaftspotential,or 2001; Rubia et al., 2001; Ullsperger & von Cramon, 2001; electrical readiness potential that can be recorded from Hazeltine, Poldrack, & Gabrieli, 2000; Kiehl, Liddle, & the scalp, that precedes preinstructed motor acts and

414 Journal of Cognitive Neuroscience Volume 15, Number 3

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 has been shown to be more prominent prior to correct unknown. Again, the SEF may be the primary oculomo- antisaccades than before prosaccades and reflexive pro- tor region that actually inhibits the reflexive prosaccade saccade errors on antisaccade trials (Everling, Spante- to the stimulus, especially given its increased neuronal kow, Krappmann, & Flohr, 1998; Everling, Krappmann, firing rate just prior to antisaccades (Schlag-Rey et al., & Flohr, 1997). 1997) and its projections to omnipause neurons in the The only significant difference in FEF and IPS brain brain stem (Huerta & Kaas, 1990; Shook, Schlagrey, & activity was limited to the stimulus-response phase of Schlag, 1988) that might mediate halting the planned the task, where correct antisaccades showed greater prosaccade until the antisaccade can be generated. activity than prosaccades and reflexive errors on anti- Because we were able to separate preparatory activity Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 saccade trials. These data are more difficult to interpret during the delay from the motor activity at response, we in the context of the current design because the primary are able to propose that the pre-SMA establishes a effect was found at the stimulus-response interval of the preparatory oculomotor set that influences how one will task, when many perceptual and motor operations are respond to the abrupt presentation of a visual stimulus. presumably co-occurring rapidly. Nonetheless, the More specifically, the pre-SMA activity reflects a highly increased activity in the FEF and IPS noted here could flexible, abstract, and directionally undetermined eye- be related to processes including covert orienting to the movement goal that biases the activity in other oculo- visual stimulus (Nobre, Gitelman, Dias, & Mesulam, motor centers, such as the FEF and/or SEF, and reduces 2000; Corbetta et al., 1998), cancellation of the reflexive the likelihood that a reflex-like saccade to the exogenous saccade (Hanes et al., 1998), and/or selection of the cue will be generated. This could be achieved through antisaccade location (Schall & Hanes, 1993). the excitation of fixation-related neurons in the FEF but Saccades are produced when activity in FEF move- is more likely accomplished through the inhibition of ment-related neurons that drive the eyes to a stimulus movement-related neurons (Everling & Munoz, 2000). increases and activity in fixation-related neurons that Although in the monkey, at least, there is no evidence lock gaze in place decreases (Everling & Munoz, 2000; for monosynaptic projections from the pre-SMA to the Hanes & Schall, 1996). At the time when the peripheral FEF, the pre-SMA shares diffuse reciprocal connections visual stimulus appears, competition between gaze- with the SEF and the dorsolateral prefrontal cortex holding and gaze-shifting mechanisms in the FEF deter- (areas 9 and 46), both of which are highly intercon- mines, through its efferent projections to the SC, nected with the FEF (Bates & Goldman-Rakic, 1993; whether the reflexive saccade is triggered or not. If Luppino et al., 1993; Schall, Morel, & Kaas, 1993; Huerta activity in movement-related neurons can be kept below & Kaas, 1990). Thus, activity in the pre-SMA could readily a critical threshold just long enough for the voluntary exert modulatory control over drive related oculomotor antisaccade to be programmed and initiated, then the neurons elsewhere, either in cortex or subcortex, pre- decision to make a correct antisaccade is likely to be venting unwanted glances. Our findings illustrate, in achieved. It is known that saccade-related neurons in general, a mechanism by which top-down controlling the SC and FEF decrease their rate of firing before signals can bias bottom-up neural processes to allow for antisaccades compared with prosaccades (Everling & willful adaptive behavior (Miller & Cohen, 2001). This Munoz, 2000; Everling et al., 1999; Everling, Dorris, interplay, between voluntary top-down and reflexive et al., 1998), but neurons in the SEF increase their rate bottom-up processes, is one of the most general organ- of firing (Schlag-Rey et al., 1997). These differences start izing principles of brain function. to emerge, however, only about 300 msec before the appearance of the saccade stimulus and would likely METHODS contribute statistically to the response period, not to the delay period, covariate. Other studies that are specifi- Participants and Experimental Methods cally designed to better address differential contribu- Twelve healthy participants (five women; ages 21–33), tions of the SEF and FEF near the response are currently who gave informed consent according to procedures underway. In this study though, the greater SEF activity approved by the University of California, performed 80 during antisaccades compared with prosaccades may antisaccade, 40 fixation, and 32 prosaccade trials in a reflect neural events just before the saccade stimulus randomly interleaved order as depicted in Figure 1. in addition to active fixation, saccade selection, and saccade production processes. The key difference Neuroimaging Methods between the pre-SMA and SEF was that the pre-SMA differentiated antisaccades and prosaccades earlier than Functional images were acquired during eight runs did the SEF, as early as 4–6 sec before the saccade lasting 418 sec each, resulting in 1672 volumes total stimulus even appeared. This signifies a primary role for covering the dorsal cortex. T2*-weighted echo-planar the pre-SMA in preparing the oculomotor system for the images (EPI) sensitive to blood oxygenation level- impending need to inhibit the reflexive saccade, even dependent (BOLD) contrasts were acquired at 4 T with when the direction of the visual stimulus and saccade is a Varian INOVA MR scanner (http://www.varianinc.com)

Curtis and D’Esposito 415

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 and a TEM send-and-receive RF head coil (http:// (200 msec) to the left or right hemifield. The HRF used www.highfieldcoils.com) using a two-shot gradient- was derived from FEF and did not differ in shape from echo–EPI sequence (22.4 cm2 field of view with a 64 Â HRFs derived from SEF. Our methods for analyzing 64 matrix size resulting in an in-plane resolution of 3.5 Â temporal patterns of brain activity (i.e., BOLD) within 3.5 mm for each of eighteen 5-mm axial slices with a trial are described in detail elsewhere (Postle, Zarahn, 0.5-mm interslice gap; repetition time = 1 sec per half & D’Esposito, 2000). Briefly, we modeled fMRI signal of k-space (2 s total), echo time = 28 msec, flip angle = changes evoked by each epoch of the trial with a 208. High-resolution MP-Flash 3D T1-weighted scans covariate shaped like the HRF by convolving it with were acquired for anatomical localization. each independent variable (instructional cue, prepara- Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 tory delay, and saccade response) (Zarahn, Aguirre, & D’Esposito, 1997) and entering the result into the Oculomotor Recording Methods modified general linear model (Worsley & Friston, Eye position was monitored in the scanner at 60 Hz with 1995) for analysis using VoxBo (http://www.voxbo.org). an infrared videographic camera equipped with a tele- On average, 24±12 SEF, 33±17 pre-SMA, 91±15 bilat- photo lens (Model 504LRO, Applied Sciences Laborato- eral FEF, and 76±22 bilateral IPS voxels were identified ries, http://www.a-s-l.com) that focused on the right eye for each participant that showed significant task-related viewed from a small dielectric flat surface mirror activity independent of trial type ( p < .05, corrected for mounted inside the RF coil. Nine-point calibrations were multiple comparisons). From these voxels, parameter performed at the beginning of the session and between estimates reflecting the percent MR signal changes runs when necessary. Eye-movement data were scored relative to baseline were estimated for each covariate offline with display routines written in MATLAB (http:// by trial type. Group random effects analyses on these www.mathworks.com). parameter estimates were performed, where significant Trial Type (prosaccade, antisaccade correct, antisaccade error) Â Trial Period (cue, delay, response) interactions Regions of Interest were followed-up with t tests ( p < .008, when a = .05 is Given the close proximity of brain regions of interest corrected for multiple comparisons). In addition to and the loss of spatial resolution inherent in spatial region-of-interest analyses, group-level random effects normalization (Brett, Johnsrude, & Owen, 2002), we analyses were performed on the entire volumes col- performed our analyses within specific regions of inter- lected. This was done to aid in the visualization of the est. The locations of the SEF, pre-SMA, FEF, and IPS data and to aid in relating our data to other studies. were derived from a two-step process. First, sulcal Statistical parametric maps (t statistics) of key contrasts anatomical landmarks in the subject’s native space were were generated for the group after the individual sub- used to define SEF (in and around the paracentral sulcus ject data were spatially normalized into standard atlas of the dorsomedial wall that did not extend rostral past space (Montreal Neurological Institute reference brain) VAC plane or ventral into cingulate sulcus), pre-SMA using routines from SPM99 (http://www.fil.ion.ucl.ac.uk/ (dorsomedial cortical wall just rostral to the VAC plane spm), resampled to 2 mm isotropic voxels, and spatially above the cingulate sulcus), FEF (extending laterally smoothed with a two-voxel Gaussian kernel. along the precentral sulcus of the dorsolateral frontal cortex beginning at the junction with the superior frontal sulcus), and the IPS (lateral sulcus dividing the Acknowledgments superior and inferior lobules in parietal cortices) in We thank Ben Inglis and Christopher Hirsch for technical accord with other studies (Grosbras, Lobel, Van de assistance and Charan Ranganath and Eric Schumacher for Moortele, LeBihan, & Berthoz, 1999; Luna et al., 1998). helpful comments. This work was supported by grants from Second, these regions were then functionally tested with NIH and from the James S. McDonnell Foundation (C.E.C.). a visually guided saccade task (see below), which in all Reprint requests should be sent to Clayton E. Curtis, Helen cases the FEF, SEF, and IPS confirmed the presence of Wills Neuroscience Institute and Department of Psychology, significant ( p < .05, corrected for multiple comparsons) Henry H. Wheeler Jr. Brain Imaging Center, University of California, 3210 Tolman Hall, Berkeley, CA 94720-1650, USA, or saccade-related activity within the anatomically defined via e-mail: [email protected]. regions; significant activity in the pre-SMA was less consistent and/or lower in magnitude across subjects. The data reported in this experiment have been deposited in The fMRI Data Center (http://www.fmridc.org). The accession number is 2-2003-113E7. Data Analysis For all participants, a hemodynamic response function REFERENCES (HRF) was empirically derived (Aguirre, Zarahn, & Aguirre, G. K., Zarahn, E., & D’Esposito, M. (1998). The D’Esposito, 1998) in response to 20 saccades made to variability of human, BOLD hemodynamic responses. flickering checkerboards (20 Hz) briefly presented Neuroimage, 8, 360–369.

416 Journal of Cognitive Neuroscience Volume 15, Number 3

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 Bates, J. F., & Goldman-Rakic, P. S. (1993). Prefrontal Hanes, D. P., Patterson, W. F., II, & Schall, J. D. (1998). Role of connections of medial motor areas in the rhesus monkey. frontal eye fields in countermanding saccades: Visual, Journal of Comparative Neurology, 336, 211–228. movement, and fixation activity. Journal of Neurophysiology, Brett, M., Johnsrude, I. S., & Owen, A. M. (2002). The problem 79, 817–834. of functional localization in the human brain. Nature Hanes, D. P., & Schall, J. D. (1996). Neural control of voluntary Reviews Neuroscience, 3, 243–249. movement initiation. Science, 274, 427–430. Bruce, C. J., Goldberg, M. E., Bushnell, M. C., & Stanton, G. B. Hazeltine, E., Poldrack, R., & Gabrieli, J. D. (2000). Neural (1985). Primate frontal eye fields. II. Physiological and activation during response competition. Journal of Cognitive anatomical correlates of electrically evoked eye movements. Neuroscience, 12, 118–129. Journal of Neurophysiology, 54, 714–734. Heide, W., Binkofski, F., Seitz, R. J., Posse, S., Nitschke, M. F.,

Connolly, J. D., Goodale, M. A., Desouza, J. F., Menon, R. S., Freund, H. J., & Kompf, D. (2001). Activation of fronto- Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 & Vilis, T. (2000). A comparison of frontoparietal fMRI parietal cortices during memorized triple-step sequences of activation during anti-saccades and anti-pointing. Journal saccadic eye movements: An fMRI study. European Journal of Neurophysiology, 84, 1645–1655. of Neuroscience, 13, 1177–1189. Corbetta, M., Akbudak, E., Conturo, T. E., Snyder, A. Z., Hikosaka, O., Sakai, K., Miyauchi, S., Takino, R., Sasaki, Y., & Ollinger, J. M., Drury, H. A., Linenweber, M. R., Petersen, Putz, B. (1996). Activation of human presupplementary S. E., Raichle, M. E., Van Essen, D. C., & Shulman, G. L. motor area in learning of sequential procedures: A functional (1998). A common network of functional areas for attention MRI study. Journal of Neurophysiology, 76, 617–621. and eye movements. Neuron, 21, 761–773. Huerta, M. F., & Kaas, J. H. (1990). Supplementary eye field Curtis, C. E., & D’Esposito, M. (2001). Dorsal frontal cortical as defined by intracortical microstimulation—connections contributions to inhibitory control: fMRI evidence from in macaques. Journal of Comparative Neurology, 293, an antisaccade task. Society for Neuroscience Abstracts, 299–330. 27, 782.6. Humberstone, M., Sawle, G. V., Clare, S., Hykin, J., Coxon, R., Deiber, M. P., Honda, M., Ibanez, V., Sadato, N., & Hallett, M. Bowtell, R., Macdonald, I. A., & Morris, P. G. (1997). (1999). Mesial motor areas in self-initiated versus externally Functional magnetic resonance imaging of single motor triggered movements examined with fMRI: Effect of events reveals human presupplementary motor area. movement type and rate. Journal of Neurophysiology, Annals of Neurology, 42, 632–637. 81, 3065–3077. Kawashima, R., Tanji, J., Okada, K., Sugiura, M., Sato, K., Dorris, M. C., Pare, M., & Munoz, D. P. (1997). Neuronal activity Kinomura, S., Inoue, K., Ogawa, A., & Fukuda, H. (1998). in monkey superior colliculus related to the initiation of Oculomotor sequence learning: A positron emission saccadic eye movements. Journal of Neuroscience, 17, tomography study. Experimental Brain Research, 122, 1–8. 8566–8579. Kiehl, K. A., Liddle, P. F., & Hopfinger, J. B. (2000). Error Everling, S., Dorris, M. C., Klein, R. M., & Munoz, D. P. (1999). processing and the rostral anterior cingulate: An event- Role of primate superior colliculus in preparation and related fMRI study. Psychophysiology, 37, 216–223. execution of anti-saccades and pro-saccades. Journal Kimmig, H., Greenlee, M. W., Gondan, M., Schira, M., of Neuroscience, 19, 2740–2754. Kassubek, J., & Mergner, T. (2001). Relationship between Everling, S., Dorris, M. C., & Munoz, D. P. (1998). Reflex saccadic eye movements and cortical activity as measured suppression in the anti-saccade task is dependent on by fMRI: Quantitative and qualitative aspects. Experimental prestimulus neural processes. Journal of Neurophysiology, Brain Research, 141, 184–194. 80, 1584–1589. Kurata, K., Tsuji, T., Naraki, S., Seino, M., & Abe, Y. (2000). Everling, S., Krappmann, P., & Flohr, H. (1997). Cortical Activation of the dorsal premotor cortex and pre- potentials preceding pro- and antisaccades in man. Electro- supplementary motor area of humans during an auditory encephalography and Clinical Neurology, 102, 356–362. conditional motor task. Journal of Neurophysiology, 84, Everling, S., & Munoz, D. P. (2000). Neuronal correlates for 1667–1672. preparatory set associated with pro-saccades and anti- Lu, M. T., Preston, J. B., & Strick, P. L. (1994). Interconnections saccades in the primate frontal eye field. Journal of between the prefrontal cortex and the premotor areas in Neuroscience, 20, 387–400. the . Journal of Comparative Neurology, 341, Everling, S., Spantekow, A., Krappmann, P., & Flohr, H. 375–392. (1998). Event-related potentials associated with correct Luna, B., Thulborn, K. R., Strojwas, M. H., McCurtain, B. J., and incorrect responses in a cued antisaccade task. Berman, R. A., Genovese, C. R., & Sweeney, J. A. (1998). Experimental Brain Research, 118, 27–34. Dorsal cortical regions subserving visually guided saccades Garavan, H., Ross, T. J., & Stein, E. A. (1999). Right hemispheric in humans: An fMRI study. Cerebral Cortex, 8, 40–47. dominance of inhibitory control: An event-related functional Luppino, G., Matelli, M., Camarda, R., & Rizzolatti, G. (1993). MRI study. Proceedings of the National Academy of Corticocortical connections of area F3 (SMA-proper) and Sciences, U.S.A., 96, 8301–8306. area F6 (pre-SMA) in the macaque monkey. Journal of Gottlieb, J., & Goldberg, M. E. (1999). Activity of neurons in Comparative Neurology, 338, 114–140. the lateral intraparietal area of the monkey during an Menon, V., Adleman, N. E., White, C. D., Glover, G. H., & Reiss, antisaccade task. Nature Neuroscience, 2, 906–912. A. L. (2001). Error-related brain activation during a Go/NoGo Grosbras, M. H., Leonards, U., Lobel, E., Poline, J. B., response inhibition task. Human Brain Mapping, 12, LeBihan, D., & Berthoz, A. (2001). Human cortical 131–143. networks for new and familiar sequences of saccades. Merriam, E. P., Colby, C. L., Thulborn, K. R., Luna, B., Olson, Cerebral Cortex, 11, 936–945. C. R., & Sweeney, J. A. (2001). Stimulus-response incompat- Grosbras, M. H., Lobel, E., Van de Moortele, P. F., LeBihan, D., ibility activates cortex proximate to three eye fields. & Berthoz, A. (1999). An anatomical landmark for the Neuroimage, 13, 794–800. supplementary eye fields in human revealed with functional Miller, E. K., & Cohen, J. D. (2001). An integrative theory of magnetic resonance imaging. Cerebral Cortex, 9, 705–711. prefrontal cortex function. Annual Review of Neuroscience, Hallett, P. (1978). Primary and secondary saccades to goals 24, 167–202. defined by instructions. Vision Research, 18, 1279–1296. Nobre, A. C., Gitelman, D. R., Dias, E. C., & Mesulam, M. M.

Curtis and D’Esposito 417

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021 (2000). Covert visual spatial orienting and saccades: Schall, J. D., Morel, A., & Kaas, J. H. (1993). Topography of Overlapping neural systems. Neuroimage, 11, 210–216. supplementary eye field afferents to frontal eye field in O’Driscoll, G. A., Alpert, N. M., Matthysse, S. W., Levy, D. L., macaque: Implications for mapping between saccade Rauch, S. L., & Holzman, P. S. (1995). Functional coordinate systems. Visual Neuroscience, 10, 385–393. neuroanatomy of antisaccade eye movements investigated Schiller, P. H., Sandell, J. H., & Maunsell, J. H. (1987). The with positron emission tomography. Proceedings of the effect of frontal eye field and superior colliculus lesions National Academy of Sciences, U.S.A., 92, 925–929. on saccadic latencies in the rhesus monkey. Journal of Paus, T. (1996). Location and function of the human frontal Neurophysiology, 57, 1033–1049. eye-field: A selective review. Neuropsychologia, 34, Schlag, J., & Schlag-Rey, M. (1987). Evidence for a supple- 475–483. mentary eye field. Journal of Neurophysiology, 57, 179–200.

Petit, L., Courtney, S. M., Ungerleider, L. G., & Haxby, J. V. Schlag-Rey, M., Amador, N., Sanchez, H., & Schlag, J. (1997). Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/15/3/409/1757819/089892903321593126.pdf by guest on 18 May 2021 (1998). Sustained activity in the medial wall during working Antisaccade performance predicted by neuronal activity in memory delays. Journal of Neuroscience, 18, 9429–9437. the supplementary eye field. Nature, 390, 398–401. Picard, N., & Strick, P. L. (2001). Imaging the premotor areas. Shima, K., Mushiake, H., Saito, N., & Tanji, J. (1996). Role for Current Opinions in Neurobiology, 11, 663–672. cells in the presupplementary motor area in updating motor Postle, B. R., Zarahn, E., & D’Esposito, M. (2000). Using plans. Proceedings of the National Academy of Sciences, event-related fMRI to assess delay-period activity during U.S.A., 93, 8694–8698. performance of spatial and nonspatial working memory Shook, B. L., Schlagrey, M., & Schlag, J. (1988). Direct projection tasks. Brain Research. Brain Research Protocols, 5, 57–66. from the supplementary eye field to the nucleus raphe Rizzolatti, G., & Luppino, G. (2001). The cortical motor system. interpositus. Experimental Brain Research, 73, 215–218. Neuron, 31, 889–901. Sweeney, J. A., Mintun, M. A., Kwee, S., Wiseman, M. B., Brown, Rubia, K., Russell, T., Overmeyer, S., Brammer, M. J., Bullmore, D. L., Rosenberg, D. R., & Carl, J. R. (1996). Positron emission E. T., Sharma, T., Simmons, A., Williams, S. C., Giampietro, V., tomography study of voluntary saccadic eye movements Andrew, C. M., & Taylor, E. (2001). Mapping motor inhibition: and spatial working memory. Journal of Neurophysiology, Conjunctive brain activations across different versions of go/ 75, 454–468. no-go and stop tasks. Neuroimage, 13, 250–261. Ullsperger, M., & von Cramon, D. Y. (2001). Subprocesses of Rushworth, M. F., Hadland, K. A., Paus, T., & Sipila, P. K. performance monitoring: A dissociation of error processing (2002). Role of the human medial frontal cortex in task and response competition revealed by event-related fMRI switching: A combined fMRI and TMS study. Journal of and ERPs. Neuroimage, 14, 1387–1401. Neurophysiology, 87, 2577–2592. Worsley, K. J., & Friston, K. J. (1995). Analysis of fMRI Sakai, K., Hikosaka, O., Miyauchi, S., Sasaki, Y., Fujimaki, N., & time-series revisited—again. Neuroimage, 2, 173–181. Putz, B. (1999). Presupplementary motor area activation Yantis, S., & Jonides, J. (1984). Abrupt visual onsets and during sequence learning reflects visuo-motor association. selective attention: Evidence from visual search. Journal Journal of Neuroscience, 19, 1–6. of Experimental Psychology: Human Perception and Schall, J. D., & Hanes, D. P. (1993). Neural basis of saccade Performance, 10, 601–621. target selection in frontal eye field during visual search. Zarahn, E., Aguirre, G., & D’Esposito, M. (1997). A trial-based Nature, 366, 467–469. experimental design for fMRI. Neuroimage, 6, 122–138.

418 Journal of Cognitive Neuroscience Volume 15, Number 3

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892903321593126 by guest on 27 September 2021